Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Under Peer Review · Attention SAE Research Paper
    3. GPT2-Small
    4. Attention Out
    5. 3-ATT-KK
    6. 7803
    Prev
    Next
    INDEX
    Explanations

    attends to proper nouns or specific terms from subsequent mentions or variations of the same word.

    oai_attention-head · claude-3-5-sonnet-20240620Triggered by @danielepoterti

    references to specific names and entities, particularly related to people, locations, and environmental terms

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 3-att-kk
    Configuration
    ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L3_Hcat_z_lr1.20e-03_l19.00e-01_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    Data Type
    float32
    Hook Name
    blocks.3.attn.hook_z
    Hook Layer
    3
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Head Attr Weights
    0:0.61
    1:0.02
    2:0.03
    3:0.02
    4:0.07
    5:0.02
    6:0.05
    7:0.03
    8:0.02
    9:0.01
    10:0.04
    11:0.02
    Negative Logits
     pause
    -1.69
     leap
    -1.60
     MIC
    -1.59
     Warn
    -1.58
     Ce
    -1.55
     deadlines
    -1.54
     BUS
    -1.53
     STA
    -1.52
     EUR
    -1.52
     LU
    -1.52
    POSITIVE LOGITS
    burgh
    2.03
    Frameworks
    2.02
    velt
    1.90
    igl
    1.81
    idy
    1.80
     mathemat
    1.77
    luaj
    1.76
    anyahu
    1.72
    gerald
    1.69
    ollen
    1.69
    Activations Density 0.595%

    No Known Activations